Bardi Verb Morphology in Historical Perspective

Claire Bowern
PhD Dissertation, Harvard University, 2004
Dissertation director: Prof. Jay Jasanoff

Many languages in Northern Australia exhibit complex verbal predication. Most commonly a
subset of inflecting verbs appears with an uninflecting ‘coverb’; the two constituents jointly
determine the argument structure and Aktionsart of the predicate. My dissertation is an
investigation of these issues in Bardi (an endangered Nyulnyulan language spoken on the North-
Western Australian coast) and the languages to which it is related. I examine possible synchronic
analyses and reconstruct the history of the formation of the systems between Proto-Nyulnyulan
and the modern attested languages. There has been very little previous work on the history of
complex predicates, and no detailed historical reconstruction for the Nyulnyulan family.
Synchronically, there are issues in the analysis of predicate structure that reflect fundamental
assumptions about the nature of generative grammar, such as lexical “adicity” (the ability of
lexical items to be modified by their syntactic context) and the role of morphology and syntax (if
there is one) in the lexicon. More basically: what should the analysis of a predicate with simple
event structure but complex argument structure be? Various analyses have been proposed for
North Australian languages, ranging from ‘compounds’ to ‘verb + auxiliary’ to ‘verb +
classifier’ to complex predicates. I present evidence that the structures in Bardi are true complex
predicates and that auxiliation, compounding, or ‘classifier’ analyses cannot account for all the
data.
Diachronically, too, there are many intriguing problems in the syntax of Nyulnyulan
languages that warrant investigation. Although the two branches of the family are very close, the
number and type of simple predicates in each branch is very different. Why, for example, should
so few inflecting verb roots be cognate between Eastern and Western Nyulnyulan when the
languages otherwise share many lexical cognates? Why do Western Nyulnyulan languages have
double the number of verb roots that the Eastern languages do? The answers to such questions
can be found in the etymology of the inflecting roots themselves. In the modern Nyulnyulan
languages the simple inflecting verb roots are monomorphemic. Historically, however, many
such roots appear to include an ‘incorporated’ element (either phrasal incorporation or argument
incorporation, or perhaps compounding). It is striking that the verbal elements used in root
formation of this type are the same light verbs that are used in complex predicates in the modern
Nyulnyulan languages. The complex predicates are also very interesting from the point of view
of syntactic reconstruction, and the tracing of calquing and language contact. Closed class simple
predicates and (co)verb + verb complex predicates are an areal feature of Australia’s North-West
and cross-cut genetic families. Thus a further result of my dissertation is the elucidation of the
relative roles of calquing, lexical borrowing and internal change that have caused the
diversification of the Nyulnyulan languages. This is especially apposite in the current context of
debates on the place of genetic linguistics in Australia and the utility of reconstruction methods
developed for Indo-European languages. I show that the same methods apply to Australian
languages.
 This dissertation falls into two parts. The first is introductory. Because the Nyulnyulan
languages and the area in which they are spoken are not widely known in the historical and
syntactic literature, I give information on the speakers and their history, and a brief sketch of the
non-verbal information important for an understanding of Bardi clause structure. I also provide
an overview of the major points of the phonological system which Nyulnyulan languages share,
and discuss the sound changes which the various languages (and Bardi in particular) have
undergone.
The second part contains the synchronic and diachronic analyses of Bardi’s simple and
complex predicate structures. Inflecting verb roots in Bardi are synchronically unanalyzable;
historically however they show relics of incorporation. I show in Chapter 5 that Proto-
Nyulnyulan is most likely to have had a syntactic process of noun incorporation which was
preserved longer in the Western Nyulnyulan languages than in the Eastern languages. This
accounts for the number of verb roots which can be reconstructed only to one branch of the
family. In chapters 6-8 I discuss the affixal morphology which inflecting roots may exhibit. The
Nyulnyulan languages have complex verbal agreement paradigms and a combination of
morphological and phonological changes has made the segmentation of morphemes an
interesting synchronic problem. I also provide a rare case study of morphological reanalysis and
restructuring in a complex morphological system. The remainder of the work is devoted to a
discussion of complex predication and verb classification. Previous work on Nyulnyulan
complex predicates has paid little attention to the syntax of these constructions and even less to a
detailed study of their historical origins. I show that despite objections by McGregor (2002) the
Bardi coverb + inflecting verb construction should, in fact, be classed as a complex predicate and
that this analysis is compatible with the functional analysis of the light verb as a verbal classifier.
In the Nyulnyulan languages we have a laboratory to test some of the predictions of theories of
complex predicate structure, and I do this in the context of previous work by Wilson (1999),
Massam (2002), and Hale and Keyser (2003).